Cervical radiculopathy is a constellation of symptoms, often including arm pain, numbness, weakness, and neck pain. Nonoperative treatment of cervical radiculopathy often incorporates various modalities, including analgesics, physical therapy, traction, orthoses, and epidural or transforaminal cervical nerve root injections1. There have been numerous reports emphasizing the efficacy of nerve root injections in delaying and/or preventing the need for surgery in both the cervical and the lumbar spine2-7. Some investigators have claimed good outcomes in as many as 80% of patients8,9. Although there have been retrospective studies of the safety of these injections, data proving the efficacy and safety of cervical nerve root injections are limited10.
Given the proximity of the vertebral artery and the robust cervical radicular arteries adjacent to the cervical nerve roots, concerns have been raised about the potential for adverse outcomes from inadvertent intravascular injections. Serious complications that may arise from inadvertent injections include cortical blindness, quadriplegia, and death secondary to infarction of the spinal cord or brainstem11,12. Tiso et al.13 examined various steroids with light microscopy and noted that particulate steroids have a tendency to aggregate, forming large crystals that are more likely than non-particulate steroids to embolize. This embolization has been theorized to be a potential mechanism for the adverse outcomes seen with particulate steroids.
We are not aware of any study in which the investigators specifically examined the potentially devastating outcomes that may result from transforaminal injection of steroids directly into the cerebrovascular system. The purpose of this study was to compare the effects of particulate and non-particulate steroid injections, into the vertebral artery, on the spinal cord and central nervous system. It was designed to be an exploratory pilot study of the use of steroids for cervical nerve root injections.
After institutional board review and approval by the university animal studies committee, we obtained eleven adult pigs (York-Landrace crossbreed) for the purposes of this study. All pigs were male and weighed between 50 and 62 kg. None of the pigs had undergone any prior medical intervention.
Surgical Technique
The pigs were premedicated with atropine (0.04 mg/kg given intramuscularly) and a cocktail consisting of Telazol (tiletamine and zolazepam), ketamine, and xylazine (1 mL/50 lb [22.7 kg] of body weight given intramuscularly). They were transported to the animal preparation room, where anesthesia was induced with use of isoflurane (1% to 5%) delivered by mask. The pigs were intubated, an intravenous catheter was placed in the marginal ear vein, and prophylactic antibiotics (15 mg/kg of cefazolin) were administered intramuscularly. The groin was shaved free of hair and prepared with use of Betadine (povidone-iodine) scrub and sterile water. The femoral artery was accessed with use of a vascular cut-down technique. A longitudinal skin incision was made, and the femoral vessels were identified by blunt dissection. Silk sutures were then placed around the femoral artery. An appropriately sized catheter sheath was placed in the artery and secured with use of the silk ties. Contrast angiography with use of Oxilan (ioxilan; 2 to 4 mL/kg/hr given intravenously) was used to identify the vertebral artery under live fluoroscopy. Appropriately sized guidewires and catheters were used to monitor the path toward the vertebral artery.
Once the contrast angiography confirmed the localization of the vertebral artery, the animals were injected with one of three drugs. In Group 1, four pigs were injected with 1 mL of methylprednisolone acetate (40 mg/mL). In Group 2, four pigs were injected with 1 mL of dexamethasone sodium phosphate, with two pigs receiving a low dose (4 mg/mL) and two pigs receiving a high dose (10 mg/mL). In Group 3, three pigs were injected with 1 mL of prednisolone sodium succinate, with one pig receiving a low dose (10 mg/mL) and two receiving a high dose (50 mg/mL).
Following injection, the animals were allowed to recover from the anesthesia, if possible, and were taken back to their cages. Animals that could not be weaned from the ventilator were subsequently killed, after magnetic resonance imaging was performed.
All animals that recovered from the injections were monitored closely until full recovery was confirmed. They were then housed overnight.
Technique and Protocol for Magnetic Resonance Imaging
In Groups 2 and 3, one pig treated with each dose concentration (high and low) underwent magnetic resonance imaging twenty-four hours following injection. Two animals in Group 1, which did not recover from the injections, also underwent magnetic resonance imaging, approximately four hours following injection. The animals were sedated and supported by a ventilator prior to transfer to the magnetic resonance imaging facilities. Throughout the process they were kept sedated with the use of isoflurane (1% to 5% in a 1:2 oxygen-nitrous oxide mixture).
Magnetic resonance imaging was carried out with a MAGNETOM Tim Trio 3.0 scanner (Siemens Medical Solutions, Malvern, Pennsylvania) with use of a twelve-channel head array coil. A three-plane localizer was used. Sagittal scans were obtained with use of an MP-RAGE (magnetization-prepared rapid gradient-echo) isotropic 0.9-mm protocol, with a T2-weighted SPACE (sampling perfection with application optimized contrasts) three-dimensional sequence with isotropic 0.9-mm spatial resolution. Axial images were obtained with FLAIR (fluid attenuated inversion recovery) transverse imaging (slice thickness, 2 mm; repetition time, 9000 msec; echo time, 94 msec). Diffusion-weighted spin-echo images were acquired with EPI (echo planar imaging) diffusion-trace-weighted scan-3 direction with b-values of 0, 500, and 1000 s/mm2 in the slice direction.
Histopathologic Evaluation
All eleven animals in the study were subsequently killed. Dissection was then performed to obtain brain and upper spinal cord neural material. Sharp dissection was carried down to the skull. A saw was used to divide the skull and the upper cervical vertebrae in order to gain access to the neural material. Sharp instrumentation was then used to remove the neural material in whole.
Following harvest of the neural material, each specimen was fixed in 4% paraformaldehyde for approximately one week. The specimens were then individually sectioned with sharp instrumentation, dehydrated in graded alcohols and xylene, embedded in paraffin, and cut into 6-µm-thick sections. Hematoxylin and eosin and periodic acid-Schiff stains were used.
Statistical Analysis
We calculated one-sided 95% confidence intervals with use of the modified Wald method for this study.
Clinical Findings
None of the four animals in Group 1 recovered from the bolus injection of 1 mL of methylprednisolone (estimate of mortality = 1.0; 95% confidence interval = 0.54 to 1.0). They all required ventilatory support after demonstrating no ability to breathe independently. Attempts to wean the animal from the ventilator led to desaturation as no spontaneous, voluntary effort to self-ventilate occurred (Table I). The two Group-2 animals that received the low dose of dexamethasone recovered from the anesthesia without complications and resumed baseline preoperative activities (estimate of mortality = 0; 95% confidence interval = 0 to 0.63). In comparison with this low-dose group, the two Group-2 animals that received the high dose of dexamethasone took 1.5 to two hours longer to show full signs of recovery. Both the high and the low-dose subsets in Group 2 regained preoperative activity levels once recovery from the procedure was complete (estimate of mortality = 0; 95% confidence interval = 0 to 0.63). All three Group-3 animals, both the one that received the low dose of prednisolone and those that received the high dose, recovered fully from the anesthesia without complications and with full recovery of preoperative activity levels (estimate of mortality = 0; 95% confidence interval = 0 to 0.53).
Findings on Magnetic Resonance Imaging
Two Group-1 animals underwent magnetic resonance imaging. Use of the specified diffusion-weighted technique mentioned above revealed distinct areas of signal change in the midbrain and brainstem of both animals (estimate of morbidity = 1.0; 95% confidence interval = 0.37 to 1.0) (Fig. 1-A, Table I). The frontal lobes did not demonstrate any notable edema.
Neither of the two animals that underwent magnetic resonance imaging in Group 2 and neither of the two that underwent it in Group 3 demonstrated any evidence of edema or distinct pathologic change (estimate of morbidity = 0; 95% confidence interval = 0 to 0.63). All aspects of the upper cervical cord and brain were well visualized and appeared to be normal (Fig. 1-B, Table I).
Histopathologic Findings
In Group 1, sections of the midbrain, medulla, and occipital lobe of all four pigs demonstrated evidence of eosinophilic neuronal necrosis (estimate of morbidity = 1.0; 95% confidence interval = 0.54 to 1.0), with condensation of neuronal nuclei, surrounded by vacuolated neurophil and white-matter pallor. There was a distinct serpentine pattern of hypodensity, vacuolation, and edema consistent with early infarction (Figs. 2-A and 2-B, Table I). These findings were in the midbrain, hippocampus, medulla, and temporal-occipital lobe.
The specimens in Groups 2 and 3 did not demonstrate any distinct patterns of infarction or eosinophilic neuronal necrosis (Figs. 2-C and 2-D, Table I).
This study demonstrated a dramatic difference between the clinical, magnetic resonance imaging, and histopathologic findings of the group treated with particulate steroids and the one treated with non-particulate steroids. Methylprednisolone is a particulate steroid that is widely used for epidural steroid injections into the cervical and lumbar spine and into joints. The particulate carrier that is used enhances the drug's half-life in order to facilitate its efficacy. However, the effects of this particulate material on the central nervous system have not been fully elucidated. In our study, none of the animals injected with methylprednisolone recovered baseline neurologic function. All four sustained a clinical cerebrovascular insult, leading to an inability to be weaned from the ventilator.
Historically, both particulate and non-particulate steroids have been used for epidural and nerve root injections. In general, steroids have numerous effects in the body. These include alterations in carbohydrate, protein, and lipid metabolism; maintenance of fluid and electrolyte balance; and preservation of normal function of the immune, musculoskeletal, and other systems14. It is the anti-inflammatory effect of steroids that is believed to account for the pain relief that may be seen with cervical nerve root injections15. Particulate steroids are delivered by means of a carrier and are thus less soluble than non-particulate steroids. Non-particulate steroids have been demonstrated to dissipate rapidly and therefore may have a limited duration of effect14. Methylprednisolone, a particulate steroid, is often the drug of choice for cervical nerve root injections as a result of its long-lasting effects. Other agents may also be used for these injections, including dexamethasone—a non-particulate steroid, triamcinolone—a particulate steroid, and betamethasone—a particulate steroid. To our knowledge, there have been no prospective, randomized controlled studies that have specifically shown an advantage of one steroid preparation over another, but the general consensus is that particulate steroids are longer-lasting2.
The non-particulate steroids used in this study were dexamethasone and prednisolone. Dexamethasone can also be used for epidural injection, and prednisolone is primarily used in animals and is administered intravenously. The concentration of drugs used in this study were based on the clinical doses of methylprednisolone. Epidural injections of methylprednisolone usually consist of a 1-mL bolus of 40 mg/mL of the drug. The dose equivalent of 4 mg of methylprednisolone is 0.75 mg of dexamethasone, which is equivalent to 5 mg of prednisolone. Dexamethasone is packaged in a low-dose form (4 mg/mL) and a high-dose form (10 mg/mL). A consistent volume of drug (1 mL) was delivered in order to avoid confounding factors that may result from bolus injection of higher volumes into the vertebral artery. The high-dose-dexamethasone group therefore received a higher dose equivalency relative to methylprednisolone (10 mg of dexamethasone is equivalent to 53 mg of methylprednisolone), but it did not experience a similar devastating clinical outcome. Prednisolone is available in 10 mg/mL and 50 mg/mL-dose concentrations. The high-dose form of this drug was essentially equivalent to the 40 mg/mL form of methylprednisolone, yet these pigs recovered with no adverse clinical manifestations. This suggests that the neurologic impact of intravascular injections is affected more by the solubility of the steroid than its dose.
Clinical examination, magnetic resonance imaging, and histologic analysis demonstrated pathologic findings consistent with a cerebrovascular insult in the particulate-steroid group. Clinically, the animals never regained consciousness. Magnetic resonance imaging showed edema in the midbrain and hindbrain. On histopathologic analysis, the cells demonstrated advanced pyknosis, cell membrane degeneration, and nuclear breakdown. Most of these changes were seen in the midbrain and brainstem. It is these regions that one would expect to be most affected, with clinical evidence of respiratory compromise. In contrast, the animals in Groups 2 and 3 recovered fully, regaining baseline activity levels and demonstrating no notable radiographic or histopathologic abnormalities.
This study may be clinically relevant for humans as transforaminal epidural cervical nerve root blocks are commonly used as part of the management of cervical radiculopathy. Prior case reports have demonstrated complications that may result from the use of these injections. Many of these adverse effects are believed to arise from injection of material into the vertebral or, more commonly, radicular arteries. These complications include cerebellar infarct, brainstem herniation, brainstem infarction, spinal cord infarction, and generalized cerebral edema. In all of these reported cases, particulate steroids (either methylprednisolone or triamcinolone) were administered with use of a transforaminal approach4,16-19. Most of the affected patients subsequently died. The model in our study was designed to allow us to examine complications of transforaminal injections. Interlaminar injections have been shown to be safer, as they avoid the radicular arteries and there is a lower chance of inadvertent penetration of the vertebral artery20. Additional clinical studies or meta-analyses could shed light onto the comparative safety and clinical efficacy of these two modes of injection.
Weaknesses of this study include the limited number of animals that were used, the limited use of magnetic resonance imaging, and the lack of direct evidence that particulate steroids led to the observed clinical findings. As this study was designed to be an exploratory pilot study, future endeavors involving more animals may provide further support of the findings. Histologically, changes consistent with edema and necrosis were noted in the methylprednisolone group; however, there was no direct evidence of damage mediated by particulate steroids. Nonetheless, as the only difference between Group 1 and Groups 2 and 3 was the presence of particulate steroids, there appears to be a strong relationship between the presence of particulate steroids and the observed clinical findings. Tiso et al.13 used electron microscopy to study various steroid formulations and noted that particulate steroids form microaggregates of at least 100 µm in size. This finding is in contrast to the 10-µm size of microaggregates of non-particulate steroids. The authors concluded that blockage of smaller vessels by the larger aggregates of particulate steroids leads to ischemic changes in the brains of affected animals. The findings of our study are consistent with this theory.
Many studies aimed at addressing the specific anatomy of the central nervous system have shown adverse outcomes following injections21,22. Fluoroscopic (contrast-enhanced) and computed-tomography-guided imaging have been introduced as techniques to ensure precise placement of needles in these vital regions5,23-25. Nonetheless, there have continued to be case reports of devastating complications during selective transforaminal cervical nerve root injections of steroids16-19,23,26.
Given the catastrophic consequences of inadvertent intravascular injection observed in this study, there should be further research on the relative efficacy of particulate compared with non-particulate steroids. Although the frequency of intravascular complications from particulate injections is low, a well-designed prospective study comparing the clinical efficacy of the two formulations would be beneficial. This will undoubtedly provide a useful risk-benefit analysis that may guide clinicians in their treatment of cervical radiculopathy. 